bpcp-package {bpcp}

Pointwise Confidence Intervals Associated with the Survival Distribution for Right Censored Data

Description

The package has functions to give several different methods for calculating pointwise confidence intervals for a single survival distribution for right censored data. There is also a two-sample test for dissimilarity (measured by difference, ratio, or odds ratio) between two survival distributions at a fixed point in time.

The recommended confidence interval for a single sample is the beta product confidence procedure (using bpcp), and the recommended test for the two-sample test is the melded BPCP test (using bpcp2samp).

Other confidence intervals and two sample tests are included in the package primarily to compare them to the recommended ones. (And justify the recommendations).

Also included is a difference in medians test that applies only to non-censored data and is designed to guarantee coverage for all sample sizes (see mdiffmedian.test). The test makes no assumptions about the distributions, so that, unlike the Hodges-Lehmann method, tied data are allowed and a shift assumption is not needed.

Details

Package:	bpcp
Type:	Package
Version:	1.4.2
Date:	2022-03-11
License:	GPL2
LazyLoad:	yes

The most important function for the single sample case is the bpcp which gives confidence intervals for the survival distribution for right censored data with nice small sample properties. That function creates an kmciLR object which has 6 methods: summary (create a data frame with estimates and confidence intervals), plot (plot Kaplan-Meier with confidence intervals), lines (add confidence intervals to a plot), StCI (pick out survival and confidence interval at specific times), median (pick out median and confidence interval on median), and quantile (pick out any quantile and its confidence interval). A mid-p option for bpcp is now available. It gives closer to nominal coverage than the standard (midp=FALSE) BPCP. For details see Fay et al (2013) on the standard BPCP and Fay and Brittain (2016) on the mid-p option.

For the two-sample test see bpcp2samp. This test for equality reduces to Fisher's exact test when there is no censoring. When there is censoring, the test is expected to maintain at least nominal coverage. For details see Fay et al (2015).

Author(s)

Michael P. Fay

Maintainer: Michael P. Fay <[e-mail]>

References

Fay, MP, Brittain, E, and Proschan, MA. (2013). Pointwise Confidence Intervals for a Survival Distribution with Small Samples or Heavy Censoring. Biostatistics 14(4): 723-736 doi: 10.1093/biostatistics/kxt016. (copy available at http://www.niaid.nih.gov/about/organization/dcr/brb/staff/Pages/michael.aspx).

Fay, MP, Proschan, MA, and Brittain, E (2015). Combining One Sample Confidence Procedures for Inference in the Two Sample Case. Biometrics 71:146-156.

Fay, MP, and Brittain, E (2016). Finite Sample Pointwise Confidence Intervals for a Survival Distribution with Right-Censored Data. Statistics in Medicine. doi: 10.1002/sim.6905.

See Also

bpcp

Examples

data(leuk)
## since there are ties at time=6
## and the data are truncated to the nearest integer, use Delta=1
bfit<-bpcp(leuk$time,leuk$status,Delta=1)
## plot Kaplan-Meier and 95 pct Beta Product Confidence Intervals 
plot(bfit,xlab="time (weeks)")
## details
summary(bfit)
quantile(bfit)
StCI(bfit,2)